Re: Patch for cygwin: silence cwrapper compilation warnings

2008-04-29 Thread Charles Wilson

Charles Wilson wrote:

2008-04-25  Charles Wilson  ...

Ensure cwrapper compiles without warnings under -std=c99:

* libltdl/config/ltmain.m4sh (func_emit_wrapper_part1):
new function.
(func_emit_wrapper_part2): new function.
(func_emit_wrapper): delegate to new functions.
(func_emit_cwrapperexe_src) [__CYGWIN__  __STRICT_ANSI__]:
ensure realpath is declared.
(func_emit_cwrapperexe_src): declare two different strings
to each hold part of the wrapper script content. Initialize
using new func_emit_wrapper_partX functions.
(func_emit_cwrapperexe_src) [main]: when emitting wrapper
script content, use both strings.



However, this bit in the posted patch, in func_emit_wrapper():
+   func_emit_wrapper_part1 ${func_emit_wrapper_arg1}
+   func_emit_wrapper_part2 ${func_emit_wrapper_arg2}
should really be
+   func_emit_wrapper_part1 ${func_emit_wrapper_arg1}
+   func_emit_wrapper_part2 ${func_emit_wrapper_arg1}
because func_emit_wrapper() only has *arg1 and should pass that argument 
on to both of the func_emit_wrapper_part*() sub-functions.


Okay for push, with the above change and rebasing?

BTW, what's the normal procedure after rebasing your topic branch, and 
pulling into (local) master [and, according to Jim's suggestions on 
gnulib mailing list, finally adding your commit info to the ChangeLog, and

  git-add ChangeLog
  git commit --amend -e
]?

On CVS, instead of 'cvs commit' you'd do 'clcommit'
On GIT, instead of 'git-push' you do...?

--
Chuck




Re: Patch for cygwin: silence cwrapper compilation warnings

2008-04-25 Thread Charles Wilson

Charles Wilson wrote:


2008-04-25  Charles Wilson  ...

Ensure cwrapper compiles without warnings under -std=c99:

* libltdl/config/ltmain.m4sh (func_emit_wrapper_part1):
new function.
(func_emit_wrapper_part2): new function.
(func_emit_wrapper): delegate to new functions.
(func_emit_cwrapperexe_src) [__CYGWIN__  __STRICT_ANSI__]:
ensure realpath is declared.
(func_emit_cwrapperexe_src): declare two different strings
to each hold part of the wrapper script content. Initialize
using new func_emit_wrapper_partX functions.
(func_emit_cwrapperexe_src) [main]: when emitting wrapper
script content, use both strings.


Test resuits:
==
All 115 tests passed
(9 tests were not run)
==

ERROR: 68 tests were run,
6 failed (3 expected failures).
3 tests were skipped.

Failed:
 10: verbatim aclocal.m4 w/o AC_CONFIG_MACRO_DIR FAILED
 23: Link order of deplibs.  FAILED
 70: Run tests with low max_cmd_len  FAILED

I believe #10 is already-reported fallout from one of Gary's patches, 
and #23 and #70 are expected on cygwin.


But see:
http://cygwin.com/ml/cygwin/2008-04/msg00681.html


However, this bit in the posted patch, in func_emit_wrapper():
+   func_emit_wrapper_part1 ${func_emit_wrapper_arg1}
+   func_emit_wrapper_part2 ${func_emit_wrapper_arg2}
should really be
+   func_emit_wrapper_part1 ${func_emit_wrapper_arg1}
+   func_emit_wrapper_part2 ${func_emit_wrapper_arg1}
because func_emit_wrapper() only has *arg1 and should pass that argument 
on to both of the func_emit_wrapper_part*() sub-functions.


--
Chuck

P.S. I think I might have to switch to dash for my libtool work. Running 
the testsuite on cygwin-1.5.25-12, Vista Home Premium 32bit, 3G Ram, 
dual core centrino mobile T5250...took 4.5 hours.